If you have wondered how scientists read gene activity one cell at a time, single cell rna sequencing explained simply is this: each cell’s RNA is tagged with a barcode, sequenced, and sorted back to its cell.
The sections below explain the steps in plain language and say what the method can and cannot tell you.
| Item | Value |
|---|---|
| What links a transcript to its cell | DNA barcodes attached to each cell’s RNA |
| Share of barcodes that may tag multiple cells (doublets) | about 5% in high-throughput methods |
| Drop-Seq capture scale (2018 review table) | about 150,000 cells per run |
| Fluidigm C1 capture scale (2018 review table) | about 1,000 cells |
Figures come from a 2018 NIH-hosted review and a later pipelines review. They can change, so check the sources before you rely on them.
Why measure gene activity cell by cell?
Traditional “bulk” RNA sequencing grinds up a tissue and reports an average. If a tissue holds many cell types, that average can hide rare cells or blur differences between them.
A brief overview of the technology puts it this way: “Traditional transcriptome, epigenome or proteome from bulk RNA/DNA samples can only capture the total level of signals from tissues/organs, which fail to distinguish individual cell variations.”
Single-cell methods keep each cell’s readout separate. This lets researchers ask which cell types are present, which genes each one is using, and how cells differ even within one type.
The approach sits on top of high-throughput sequencing machines. If you want the underlying chemistry of reading DNA and RNA, our guide to next generation sequencing explained covers that foundation.
How the method works, step by step
A typical experiment follows a similar path, although platforms differ in the details.
- Prepare a cell suspension. Tissue is broken into separate living cells, or sometimes nuclei.
- Isolate single cells. Cells are placed in tiny chambers, wells or oil droplets so that each is kept apart.
- Tag the RNA. Messenger RNA from each cell is labelled with a cell barcode.
- Copy and amplify. The RNA is copied into DNA, which is then multiplied.
- Sequence and sort. A sequencer reads everything at once, and software uses the barcodes to assign reads back to cells.
The 2018 review explains that the platforms it covers rely on DNA barcodes to index transcripts back to their individual cells of origin.
The same overview notes that PCR, a non-linear amplification process, is used in Smart-seq.
Droplet-based capture
Droplet systems such as Drop-Seq and the commercial 10x Genomics platform use microfluidic chips to isolate single cells with single beads in oil-encapsulated droplets, according to the same review. Each bead carries a cell barcode shared by all its oligos, plus a unique molecular identifier (UMI) on each oligo.
These are named only as neutral examples of the technology.
Plate and chamber methods
Other designs trap cells in small chambers. Their capture scale is lower, at about 1,000 cells for the Fluidigm C1, against about 150,000 cells per run for Drop-Seq in the review’s table.
Chamber systems can allow staining and imaging before the cell is broken open. That is a trade-off between scale and extra information about each cell.
From raw reads to cell types
The output is a large table of genes by cells. Analysts then clean it, reduce its dimensions and group similar cells into clusters.
The pipelines review describes the steps as experimental design, pre-processing and quality control, feature selection, dimensionality reduction, cell clustering and annotation, followed by downstream analysis including batch correction, trajectory inference and cell-cell communication.
- Clustering groups cells with similar gene activity.
Clusters are statistical groupings. Whether a cluster is a true cell type is a judgement that usually needs follow-up experiments.
What can go wrong: noise, zeros and doublets
Single cells hold very little RNA, so the data are noisy. The 2018 review notes that the gene-by-cell matrix “is expected to be mostly filled with zeros because of these dropout events.”
A zero can mean a gene was truly silent, or that its RNA was simply missed. Telling these apart is difficult, and it is a main reason results are treated with caution.
Doublets
Sometimes two cells share one barcode. The pipelines review says that cells with unexpectedly high counts and too many expressed genes “may represent doublets,” and that removal matters because “often ~5% of cell barcodes are tagging multiple cells.”
Tools such as scrublet, DoubletFinder and scds were developed to infer likely doublets from the data itself.
Batch effects
Samples processed on different days or machines can differ for technical reasons. Researchers sometimes need to combine datasets using batch correction methods, which themselves involve assumptions.
Single cell RNA sequencing: what it can and cannot tell you
Single-cell data can suggest which cell types exist and which genes distinguish them. It captures RNA, which is not the same as protein levels or what a cell actually does.
Cost and sample quality set practical limits too, and costs shown in the 2018 review may be out of date.
The field overlaps with RNA biology more broadly. For one application of RNA as a molecule in medicine, see our explainer on mrna technology explained. For the wider context, start with what is biotechnology.
Research findings from cell data are not a basis for personal health choices. Treatment decisions belong with a physician, who can weigh an individual’s situation.
This site explains the science only. Nothing here is medical advice, and no method described is a recommendation of any product or company.
Planning and reading a study with healthy scepticism
When you read a paper built on single-cell data, a few questions help judge how far its claims can stretch.
- How many cells were profiled, and from how many donors or animals?
- How were doublets and damaged cells filtered out?
- Were batches balanced across the groups being compared?
- Were cluster labels checked with a second method, such as staining or imaging?
Cells from only a few donors can show real differences between people that look like differences between groups. Good studies describe these choices openly, and results that rest on one dataset are usually treated as preliminary until others reproduce them.
Analysis choices matter as well.
Frequently asked questions
What is a cell barcode?
Show answer
It is a short DNA sequence attached to RNA from one cell, so software can sort reads back to the cell they came from.
Why are there so many zeros in the data?
Show answer
Each cell holds little RNA, and not every molecule is captured. The 2018 review says the matrix is expected to be mostly zeros because of dropout events.
What is a doublet?
Show answer
A doublet is a barcode that tags more than one cell. The pipelines review says often about 5% of barcodes in high-throughput methods may be affected.
Is single-cell sequencing the same as bulk RNA sequencing?
Show answer
No. Bulk sequencing reports an average across many cells, while single-cell methods keep each cell’s profile separate.
This article is for information only. It is general education about science and the drug industry, and it is not medical, treatment or investment advice.
